Fetal Region Detection using PyTorch and Soft Proposal Networks

PyTorch implementation of the paper 'Weakly Supervised Localisation for Fetal Ultrasound Images', DLMIA'18

[Project Page] [Paper] [Supp]

Abstract

This work addresses the task of detecting and localising fetal anatomical regions in 2D ultrasound images, where only image-level labels are present at training, i.e. without any localisation or segmentation information. We examine the use of convolutional neural network architectures coupled with soft proposal layers. The resulting network simultaneously performs anatomical region detection (classification) and localisation tasks. We generate a proposal map describing the attention of the network for a particular class. The network is trained on 85,500 2D fetal ultrasound images and their associated labels. Labels correspond to six anatomical regions: head, spine, thorax, abdomen, limbs, and placenta. Detection achieves an average accuracy of 90% on individual regions, and show that the proposal maps correlate well with relevant anatomical structures. This work presents itself as a powerful and essential step towards subsequent tasks such as fetal position and pose estimation, organ-specific segmentation, or image-guided navigation.

Requirements

• Cuda >= 8
• Python>=3.5
• PyTorch: pip3 install torch torchvision
• Other Packages: torchnet, spn

Usage

1. Install via pip:

   [sudo] pip install git+https://github.com/ntoussaint/fetalnav.git
   

2. Or fom source

   git clone https://github.com/ntoussaint/fetalnav.git
   cd fetalnav
   [sudo] python setup.py install
   

Notebook examples

Datasets and Transforms

Test the dataset loaders (using SimpleITK) and the different transforms

Train fetalnav

Train your own fetalnav network and experiment with hyperparameter tuning

Evaluate fetalnav

Use the network to infer the region and localise regions in the image

Supplementary material

A significant part of the training material for this study was produced using an in-house labelling tool that is available [here]:

Citation

If you use this code in your research, please cite:

@inproceedings{toussaint.dlmia.18,
    author = {Toussaint, Nicolas and Khanal, Bishesh and Sinclair, Matthew and Gomez, Alberto and Skelton, Emily and Matthew, Jacqueline and Schnabel, Julia A.},
    title = {Weakly Supervised Localisation for Fetal Ultrasound Images},
    booktitle = {Proceedings of the 4th Workshop on Deep Learning in Medical Image Analysis},
    year = {2018},
    note={\href{https://arxiv.org/abs/1808.00793}{[url]}}
}

Acknowledgement

This work was supported by the Wellcome/EPSRC Centre for Medical Eng. [WT203148/Z/16/Z], and the Wellcome IEH Award [102431]

Author

Nicolas Toussaint, PhD
School of Biomedical Engineering,
King's College London

Contact: nicolas.a.toussaint@kcl.ac.uk nicolas.toussaint@gmail.com